FaceAge: Predicting Biological Age and Cancer Prognosis from Facial Images

In a compelling fusion of computer vision and clinical oncology, Dennis Bontempi and colleagues introduce FaceAge, a deep learning model that estimates biological age from facial photographs to aid medical prognostication. Trained on over 58,000 images of healthy older adults, FaceAge was validated on cancer patients across three cohorts, demonstrating that appearing older than one’s chronological age significantly correlates with poorer survival outcomes. Integrating FaceAge into standard survival prediction models for patients receiving palliative care improved predictive accuracy (AUC increased from 0.74 to 0.80). Furthermore, FaceAge age estimates correlated with senescence-related gene expression, reinforcing its potential as a biomarker of molecular ageing. By transforming subjective visual assessments into objective, quantifiable inputs, FaceAge represents a scalable tool for clinical decision support, especially in end-of-life care planning.